Block Exogeneity in VAR

[Herbo]

Hello,

I am totally confused. I read many, many journal papers trying to understand the exact interpretation of " Exluded ALL" p-value.

If the p-value, for example, is .5 does that mean that variable can be removed from VAR and it can be used as X in VARX model?

Related question: Consider the following Granger Causality results of a VAR.

Dependent variable: DLFTA

Excluded Chi-sq df Prob.

DIFRS 0.226240 2 0.8930
DIV 1.377488 2 0.5022
DLDIV 5.078153 2 0.0789

All 6.4484 6 0.3749

In the above table why do we get DLDIV Granger cause DLFTA, but all other variables jointly do not cause DLFTA. Is this a contradiction, how do I
explain it?


Thanks in advance,

Herbert